A hybrid energy storage microgrid capacity configuration method based on double-layer optimization decomposition

Abstract

The application discloses a kind of hybrid energy storage microgrid capacity configuration methods based on double-layer optimization decomposition, comprising the following steps: based on annual wind light load data extraction typical day, construct the double-layer optimization model of capacity configuration with minimum as target full life cycle cost;Inner layer uses intelligent hybrid algorithm to optimize variational mode decomposition parameter, and the power shortage signal is decomposed, and the adaptive distribution of high and low frequency power is realized by energy and frequency joint clustering, respectively by super capacitor and battery bear;Outer layer is based on the allocation scheme, and the capacity of wind light storage is globally optimized by using optimization solver, and finally the optimal configuration scheme of economy that meets operation constraint is output.The application realizes the full life cycle economic optimal configuration of microgrid under the condition of meeting reliable operation constraint by constructing the capacity configuration optimization model of inner and outer layer coupling, combining signal decomposition parameter adaptive optimization algorithm and objectively distributing hybrid energy storage output according to energy and frequency characteristics.

Description

Technical Field

[0001] This invention belongs to the field of power system and automation technology, and particularly relates to a hybrid energy storage microgrid capacity configuration method based on two-layer optimization decomposition. Background Technology

[0002] With the deepening of the global energy structure transformation and the guidance of the strategic goal of "carbon peaking and carbon neutrality," the development and utilization of clean and renewable energy sources, represented by wind and solar power, has become a key path. However, wind and solar power have inherent intermittency and randomness, and their large-scale direct grid connection will bring significant power fluctuation challenges to system operation, affecting power supply reliability and power quality. Against this backdrop, microgrid technology, which integrates distributed power sources, energy storage systems, and local loads, has become an important solution for effectively integrating and efficiently utilizing such distributed renewable energy sources.

[0003] In microgrid systems, energy storage devices play an irreplaceable role in mitigating power fluctuations, enabling energy time-shifting, and ensuring stable system operation. Currently, a variety of energy storage technologies are available, primarily including batteries, supercapacitors, flywheel energy storage, and compressed air energy storage. Each technology has distinct characteristics: energy-type storage media such as batteries have high energy density, but their power response speed and cycle life are often limited; while power-type storage media such as supercapacitors possess excellent power density and fast response capabilities, their energy storage capacity is limited. A single energy storage technology cannot simultaneously achieve optimal performance across multiple key indicators such as energy density, power density, cycle life, response speed, and cost, thus limiting its overall effectiveness in addressing the complex and dynamic power demands of microgrids.

[0004] To address this, hybrid energy storage systems combining power-type and energy-type energy storage have emerged, aiming to synergistically meet the system's dual needs for energy storage and power support through technological complementarity. Currently, combining batteries and supercapacitors is a common form of hybrid energy storage in practice. However, how to scientifically and economically configure the capacity of hybrid energy storage systems—that is, determining the rated power and rated capacity of each type of energy storage to achieve optimal life-cycle cost while meeting system operating constraints—remains a prominent technical challenge in the field of microgrid planning and design. Existing methods still face many challenges in dealing with the uncertainty of wind and solar power output, coordinating the different techno-economic characteristics of the two types of energy storage, and achieving global optimization. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a hybrid energy storage microgrid capacity configuration method based on two-layer optimization decomposition, comprising the following steps: Acquire the annual wind and solar resource data and load data of the target area, and extract several typical daily data based on the wind and solar resource data and load data; Based on the typical daily data, a two-layer capacity configuration optimization model is constructed with the goal of minimizing the overall cost of the microgrid. The outer layer of the two-layer capacity configuration optimization model is used to optimize the installed capacity of distributed power sources and hybrid energy storage, while the inner layer is used to decompose the system power deficit signal and perform high and low frequency allocation. A hybrid algorithm combining differential evolution and whale optimization is used to optimize and solve the variational mode decomposition algorithm used for signal decomposition in the inner layer. The optimized variational mode decomposition algorithm is used to decompose the system power deficit signal, and the energy and frequency joint clustering method is used to adaptively divide the multiple modal components obtained by decomposition into high-frequency components and low-frequency components, so as to allocate them to the supercapacitor and the battery respectively. Based on the power allocation scheme, the outer capacity configuration problem of the capacity configuration two-layer optimization model is solved by an optimization solver, and the optimal capacity configuration results of each distributed power source and hybrid energy storage that meet the preset operating constraints are output.

[0006] Optionally, obtain the annual wind and solar resource data and load data for the target area, and extract several typical daily data based on the wind and solar resource data and load data, specifically including: Based on the time-series data of wind speed, light intensity and load acquired throughout the year, the data are classified by season; For each season's data, a clustering algorithm is used to determine the number of typical days for that season. The multiple candidate day data obtained from the clustering are then weighted and summed to finally obtain a single typical day data that represents the characteristics of the corresponding season.

[0007] Optionally, a two-level capacity configuration optimization model is constructed with the goal of minimizing the overall cost of the microgrid. The objective function of the two-level capacity configuration optimization model is the typical daily comprehensive cost, which includes the annual system investment and installation cost, annual operation and maintenance cost, wind and solar curtailment penalty cost, load interruption compensation cost, equipment replacement cost, and additional energy curtailment penalty cost. Furthermore, the constraints that the capacity configuration dual-layer optimization model must satisfy include: the upper limit constraints on the number or capacity of wind and solar power generation units and energy storage units, the state of charge and charging and discharging power constraints of batteries and supercapacitors, system power balance constraints, and the maximum output constraints of wind and solar power generation units.

[0008] Optionally, a hybrid algorithm combining differential evolution and whale optimization is used to optimize the variational mode decomposition algorithm used for signal decomposition in the inner layer, specifically including: Based on the preset optimization objective, the parameter population of the hybrid algorithm is initialized, where each individual represents a set of candidate penalty factors and modality decomposition layers; Based on the current parameter population, new candidate parameter combinations are generated using the mutation and crossover strategies of the hybrid algorithm. Based on the combination of candidate parameters, variational mode decomposition is performed on the system power deficit signal, and the sum of sample entropy of each mode component after decomposition is calculated. Based on the sum of the entropies of each sample obtained by calculation, the parameter population is updated by selecting a strategy to retain individuals with better fitness. Based on the preset iteration termination condition, the above process of generating candidate parameters, calculating sample entropy and updating the population is repeated to finally obtain the optimal penalty factor and mode decomposition layer number that minimizes the sum of sample entropy.

[0009] Optionally, the preset optimization objective is to minimize the sum of sample entropies; wherein, calculating the sum of sample entropies specifically involves: calculating the sample entropy of each component based on the time series of each modal component obtained by variational mode decomposition, and summing the sample entropy values ​​of all components.

[0010] Optionally, the step of generating new candidate parameter combinations using the mutation and crossover strategy of the hybrid algorithm specifically includes: Based on the current population, the first mutation vector is generated using the mutation operator of the differential evolution algorithm; Based on the best individual in the current population and the prey encirclement and spiral update strategy of the whale optimization algorithm, a second mutation vector is generated. Based on the preset crossover probability, the first mutation vector and the second mutation vector are crossed with the target individual to generate experimental individuals as new candidate parameter combinations.

[0011] Optionally, a joint energy and frequency clustering method is used to adaptively divide the multiple modal components obtained from the decomposition into high-frequency components and low-frequency components, specifically including: Calculate the average instantaneous frequency and energy characteristics of each modal component to form a joint feature vector for each component; Based on the joint feature vector of all modal components, a clustering algorithm is used to divide all modal components into two clusters; One cluster is identified as a high-frequency component and assigned to the supercapacitor; the other cluster is identified as a low-frequency component and assigned to the battery. After the component allocation is completed, the preset output signal is further modified according to the rated capacity and charging / discharging power limits of the battery and supercapacitor to obtain the actual executable charging / discharging power command.

[0012] Optionally, based on the power allocation scheme, an optimization solver is used to solve the outer capacity configuration problem of the two-layer capacity configuration optimization model, outputting the optimal capacity configuration results for each distributed power source and hybrid energy storage that satisfy preset operating constraints, specifically including: Based on the power allocation scheme, the capacity configuration bi-level optimization model is solved using an optimization solver. The total capacity of wind turbines, photovoltaic modules, batteries, and supercapacitors are used as decision variables. Under the condition of satisfying all preset constraints, the capacity configuration scheme that minimizes the comprehensive cost on a typical day is obtained. Based on the capacity configuration scheme and the power allocation scheme obtained from the solution, the wind curtailment rate and the load power shortage rate of the system under this configuration are calculated. The output of the capacity configuration scheme, the corresponding hybrid energy storage power configuration, and the wind curtailment rate and load power shortage rate are used as the optimal capacity configuration result.

[0013] On the other hand, the present invention also provides an electronic device including a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the method when executing the computing program.

[0014] On the other hand, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0015] Compared with the prior art, the present invention has the following advantages and technical effects: This invention is based on a dual-layer coupled model consisting of an outer layer optimized capacity configuration and an inner layer optimized power allocation. It solves the problems of macro-capacity planning and micro-power decomposition of hybrid energy storage systems in a unified manner, overcoming the limitations of traditional single-layer models that cannot coordinate global economic efficiency and local power matching. This achieves the optimal configuration effect in terms of overall cost over the entire life cycle while ensuring the reliability of system operation.

[0016] This invention is based on a hybrid intelligent algorithm that combines differential evolution algorithm and whale optimization algorithm. It adaptively optimizes the key parameters of variational mode decomposition, effectively overcoming the drawbacks of parameter dependence on empirical settings, improving the accuracy and adaptability of signal decomposition, and thus achieving the technical effect of more accurately extracting components that characterize the fluctuation characteristics at different time scales from the system power deficit, laying a high-quality foundation for subsequent energy storage allocation.

[0017] This invention is based on the proposed high-low frequency partitioning method of joint energy and frequency clustering. It comprehensively considers the amplitude energy and fluctuation frequency characteristics of each modal component, and realizes objective and adaptive classification of the decomposed components. This achieves the technical effect of avoiding subjective experience judgment bias and reasonably and automatically allocating low-frequency energy type power to batteries and high-frequency power type power to supercapacitors, so that the technical characteristics of the two energy storage devices can be fully utilized.

[0018] This invention uses the aforementioned precise power allocation scheme optimized by the inner layer to guide the capacity configuration optimization solution of the outer layer system, so that the capacity configuration result is closely coupled with the specific operation strategy. At the same time, it combines a fast solver to complete global optimization, thereby achieving the technical effect of significantly improving the solution efficiency and engineering practicality of the overall optimization model, and providing an effective tool for the rapid planning and design of microgrids. Attached Figure Description

[0019] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a model diagram of a microgrid system according to an embodiment of the present invention; Figure 2 This is a system capacity configuration strategy diagram according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating the high and low frequency power allocation process according to an embodiment of the present invention. Figure 4 This is a flowchart illustrating the calculation of system power deficit and renewable energy waste in an embodiment of the present invention. Figure 5 This is a typical daily curve diagram of spring, summer, autumn and winter obtained by clustering in an embodiment of the present invention; Figure 6 This is a spring VMD decomposition diagram according to an embodiment of the present invention; Figure 7 This is a summer VMD decomposition diagram according to an embodiment of the present invention; Figure 8 This is an autumn VMD decomposition diagram according to an embodiment of the present invention; Figure 9 This is a winter VMD decomposition diagram according to an embodiment of the present invention; Figure 10 The output curves of the system on typical days in spring, summer, autumn and winter are shown in the embodiments of the present invention. Figure 11 This is a graph showing the daily HESS system capacity variation curves for typical days in spring, summer, autumn, and winter according to an embodiment of the present invention. Figure 12 The above are typical daily VMD decomposition spectrum diagrams for spring, summer, autumn and winter seasons according to an embodiment of the present invention. Detailed Implementation

[0020] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0022] Example 1 like Figure 1 As shown, this embodiment provides a hybrid energy storage microgrid capacity configuration method based on two-layer optimization decomposition, including the following steps: Acquire the annual wind and solar resource data and load data of the target area, and extract several typical daily data based on the wind and solar resource data and load data; Based on the typical daily data, a two-layer capacity configuration optimization model is constructed with the goal of minimizing the overall cost of the microgrid. The outer layer of the two-layer capacity configuration optimization model is used to optimize the installed capacity of distributed power sources and hybrid energy storage, while the inner layer is used to decompose the system power deficit signal and perform high and low frequency allocation. A hybrid algorithm combining differential evolution and whale optimization is used to optimize and solve the variational mode decomposition algorithm used for signal decomposition in the inner layer. The optimized variational mode decomposition algorithm is used to decompose the system power deficit signal, and the energy and frequency joint clustering method is used to adaptively divide the multiple modal components obtained by decomposition into high-frequency components and low-frequency components, so as to allocate them to the supercapacitor and the battery respectively. Based on the power allocation scheme, the outer capacity configuration problem of the capacity configuration two-layer optimization model is solved by an optimization solver, and the optimal capacity configuration results of each distributed power source and hybrid energy storage that meet the preset operating constraints are output.

[0023] Feasible method: Obtain annual wind and solar resource data and load data for the target area, and extract several typical daily data based on the wind and solar resource data and load data, specifically including: Based on the acquired time-series data of wind speed, light intensity and load throughout the year, the data are classified by season. For the data of each season, a clustering algorithm is used to determine the number of typical days of the corresponding season, and the data of multiple candidate days obtained by clustering are weighted and summed to finally obtain a single typical day data used to characterize the characteristics of the corresponding season.

[0024] It is feasible to construct a two-level capacity configuration optimization model with the goal of minimizing the overall cost of the microgrid. The objective function of the two-level capacity configuration optimization model is the typical daily comprehensive cost, including the annual system investment and installation cost, annual operation and maintenance cost, wind and solar curtailment penalty cost, load interruption compensation cost, equipment replacement cost, and additional energy curtailment penalty cost. Furthermore, the constraints that the capacity configuration dual-layer optimization model must satisfy include: the upper limit constraints on the number or capacity of wind and solar power generation units and energy storage units, the state of charge and charging and discharging power constraints of batteries and supercapacitors, system power balance constraints, and the maximum output constraints of wind and solar power generation units.

[0025] A feasible hybrid algorithm combining differential evolution and whale optimization is used to optimize and solve the variational mode decomposition algorithm used for signal decomposition in the inner layer, specifically including: Based on a preset optimization objective, the parameter population of the hybrid algorithm is initialized, where each individual represents a set of candidate penalty factors and modal decomposition levels. Based on the current parameter population, new candidate parameter combinations are generated using the mutation and crossover strategies of the hybrid algorithm. Based on each candidate parameter combination, variational mode decomposition is performed on the system power deficit signal, and the sum of sample entropies of each modal component after decomposition is calculated. Based on the calculated sum of sample entropies, the parameter population is updated by selecting a strategy, retaining individuals with better fitness. Based on a preset iteration termination condition, the above process of generating candidate parameters, calculating sample entropies, and updating the population is repeated until the optimal penalty factor and modal decomposition level that minimizes the sum of sample entropies are obtained.

[0026] Furthermore, the preset optimization objective is to minimize the sum of sample entropies; wherein, calculating the sum of sample entropies specifically involves: calculating the sample entropy of each component based on the time series of each modal component obtained by variational mode decomposition, and summing the sample entropy values ​​of all components.

[0027] Furthermore, the generation of new candidate parameter combinations using the mutation and crossover strategy of the hybrid algorithm specifically includes: Based on the current population, a first mutation vector is generated using the mutation operator of the differential evolution algorithm; a second mutation vector is generated based on the best individual in the current population and the prey encirclement and spiral update strategy of the whale optimization algorithm; according to the preset crossover probability, the first mutation vector and the second mutation vector are crossed with the target individual to generate an experimental individual as a new candidate parameter combination.

[0028] Feasible implementation involves using a joint energy and frequency clustering method to adaptively divide the multiple modal components obtained from decomposition into high-frequency and low-frequency components, specifically including: The average instantaneous frequency and energy characteristics of each modal component are calculated to form a joint feature vector for each component. Based on the joint feature vector of all modal components, a clustering algorithm is used to divide all modal components into two clusters. One cluster is identified as a high-frequency component and assigned to the supercapacitor. The other cluster is identified as a low-frequency component and assigned to the battery. After the component allocation is completed, the preset output signal is further modified according to the rated capacity and charging / discharging power limit of the battery and supercapacitor to obtain the actual executable charging / discharging power command.

[0029] Feasible, based on the power allocation scheme, the outer capacity configuration problem of the two-layer capacity configuration optimization model is solved by an optimization solver, and the optimal capacity configuration results of each distributed power source and hybrid energy storage that satisfy preset operating constraints are output, specifically including: Based on the power allocation scheme, an optimization solver is used to solve the dual-level optimization model of the capacity configuration. Using the total capacity of wind turbines, photovoltaic modules, batteries, and supercapacitors as decision variables, and satisfying all preset constraints, the capacity configuration scheme that minimizes the comprehensive cost per typical day is obtained. Based on the solved capacity configuration scheme and the power allocation scheme, the wind curtailment rate and load deficit rate of the system under this configuration are calculated. The capacity configuration scheme, the corresponding hybrid energy storage power configuration, and the wind curtailment rate and load deficit rate are output as the optimal capacity configuration result.

[0030] On the other hand, this embodiment also provides an electronic device, including a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the method when executing the computing program.

[0031] On the other hand, this embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0032] Example 2 like Figure 2 As shown, the capacity configuration method of the hybrid energy storage microgrid in Example 1 will be further described below. Establish as follows... Figure 1 The microgrid system model, the method specifically includes the following steps: S1. Extract annual wind speed, light intensity, and load data from an industrial park in Xinjiang for case analysis, with a time resolution of 15 minutes, and summarize the information. Use the k-means algorithm to cluster the data for all four seasons throughout the year, and use the elbow method and silhouette coefficient method to determine the number of clusters. Finally, perform weighted summation on each typical day to obtain the data for typical days in spring, summer, autumn, and winter. S2. Based on the data provided in step S1, establish a two-layer VMD capacity configuration optimization model for an integrated microgrid of "wind-solar-hybrid energy storage," satisfying certain constraints, with the typical daily comprehensive cost of the system as the objective function. The hybrid energy storage consists of lithium iron phosphate batteries and supercapacitors. The outer layer of the model optimizes the capacity configuration of wind turbines, photovoltaic modules, and hybrid energy storage, while the inner layer allocates the output of the energy storage system, decomposing the hybrid energy storage output and allocating it to high and low frequencies. S3. Based on the two-layer capacity optimization model established in step S2, the parameters (α, K) of the VMD algorithm for decomposing the hybrid energy storage output are optimized and solved using the hybrid optimization algorithm DE_WOA, which combines the differential evolution algorithm and the whale optimization algorithm, and finally decomposed into multiple IMF power components. S4. Based on the high and low frequency allocation of the inner layer mentioned in steps S2 and S3, and after decomposing the hybrid energy storage output into multiple IMF power components in step S3, a method of energy and frequency joint clustering is proposed to divide the high and low frequencies, assigning the low frequency power components to the lithium iron phosphate battery and the high frequency components to the supercapacitor. S5. Based on the above steps, use CPLEX to solve the established system, obtain the typical daily comprehensive cost of the hybrid energy storage microgrid, and calculate the operation indicators such as wind and solar curtailment rate and load power shortage rate. Finally, obtain the system configuration results for typical days in the four seasons, and output the wind and solar power generation power, as well as the hybrid energy storage capacity and charging and discharging power.

[0033] Furthermore, the microgrid model established in step S2 has the following characteristics: (1) Wind power generation model: The output power of wind power is mainly related to wind speed, and the variation of wind speed follows a Weibull distribution with a probability density function. for: ; In the formula: Wind speed; For shape parameters; This is a proportional parameter.

[0034] Therefore, the output power model of wind power can be derived as follows: ; In the formula: This refers to the total installed capacity of the wind turbine units. To cut in wind speed; To cut off the wind speed; This is the rated wind speed.

[0035] (2) Photovoltaic power generation model: Photovoltaic power generation primarily occurs when sunlight shines on photovoltaic panels, causing the panels to undergo a photoelectric effect to produce electricity. Besides the capacity of the photovoltaic cells, the output power of a single photovoltaic cell is mainly affected by the local light intensity and ambient temperature.

[0036] Light intensity is assumed to follow a Beta distribution over a certain period of time. Its probability density function is... for: ; In the formula: Γ(·) is the gamma function; α and β are the shape parameters of the Beta distribution, respectively; These are the solar radiation intensity and the maximum solar radiation intensity, respectively.

[0037] Therefore, the probabilistic model for photovoltaic output power is as follows: ; In the formula: These are the output power and maximum output power of the photovoltaic system, respectively.

[0038] (3) Hybrid energy storage system model: The energy storage system (HESS) consists of lithium iron phosphate batteries and supercapacitors. The batteries are responsible for the low-frequency part of the microgrid system, while the supercapacitors are responsible for the high-frequency part of the microgrid system.

[0039] Battery model: The state of charge (SOC) of a battery is an important parameter for measuring the remaining capacity of a lithium iron phosphate battery. The SOC is determined by the remaining capacity at the previous moment and the charging and discharging power at the current moment.

[0040] A lithium iron phosphate battery is considered to be in a charging state when the following formula is met: ; When a lithium iron phosphate battery is in a charging state, the state of charge at time t is as shown in the following formula: ; A lithium iron phosphate battery is considered to be in a charging state when the following formula is met: ; When a lithium iron phosphate battery is in a discharged state, The state of charge at time t is given by the following equation: ; In the formula: σ is the self-discharge rate of the lithium iron phosphate battery; For lithium iron phosphate batteries Power at any moment; , Improve battery charging and discharging efficiency; This refers to the rated capacity of the lithium iron phosphate battery. for The battery's state of charge at all times.

[0041] Supercapacitor Model: The supercapacitor is considered to be in a charging state when the following formula is satisfied: ; When the supercapacitor is charging, The state of charge at time t is given by the following equation: ; The supercapacitor is considered to be in a discharged state when the following formula is satisfied: ; When the supercapacitor is in a discharging state, The state of charge at time t is given by the following equation: ; In the formula: σ is the self-discharge rate of the supercapacitor; For supercapacitors Power at any moment; , For capacitor charging and discharging efficiency; This refers to the rated capacitance of the supercapacitor. for The state of charge of the supercapacitor at all times.

[0042] Furthermore, the objective function for minimizing the typical daily comprehensive cost described in step S2 is as follows: ; In the formula: The annual investment and installation cost of the system; The annual operating and maintenance cost of the system; The penalty costs for abandoning wind and solar power; The load interruption cost for a typical day of the system; Cost of system replacement; This serves as a penalty for the system's additional waste of new energy resources.

[0043] ; ; In the formula: This represents the annual value of the investment cost for each member in the system; The installed capacity of each member (wind turbine, photovoltaic array, battery, supercapacitor); This indicates the unit cost of each member; This refers to the depreciation rate; This refers to the service life.

[0044] ; ; In the formula: This represents the annual operating and maintenance costs of each member of the system; This represents the unit power operation and maintenance cost of each component of the system. Among them, the supercapacitor is maintenance-free, with a maintenance cost of 0.

[0045] ; In the formula, This refers to the number of battery replacements during the system's lifespan. This refers to the unit replacement cost of the battery.

[0046] This invention decomposes and reconstructs the high and low frequencies of the system deficit, and corrects it according to the HESS rated capacity to obtain the actual output of the battery and hybrid energy storage. Therefore, during operation, load deficit and waste of renewable energy may occur. The system's excess power is calculated based on the actual output. .

[0047] ; when hour: ; when hour: ; In the formula: for Excess power during a given period; for Load interruption of the microgrid system during a given time period; The actual output of the hybrid energy storage system is obtained from steps S3 and S4.

[0048] ; In the formula: for The amount of wind and solar power curtailed during certain time periods; The penalty costs incurred for curtailing wind and solar power in the system; The penalty fee per unit for wind and solar power curtailment is 0.3 / (kWh).

[0049] ; In the formula: Representation system Load interruption amount during the time period; The unit cost for system load interruption compensation is set at 0.9 / (kWh).

[0050] ; ; In the formula, the Energy Curtrilment Rate (ECR) represents the energy curtrilment rate. This indicates the additional energy forfeiture penalty imposed on the system.

[0051] Furthermore, the various constraints described in step S2 are characterized as follows: In practical engineering applications, factors such as investment costs and site area should be considered. Therefore, this constraint is set as follows: In the formula: , , , These refer to the maximum number of wind turbines, the maximum number of photovoltaic panels, and the maximum capacity of batteries and supercapacitors.

[0052] To better utilize battery performance and extend its lifespan, constraints are imposed on the battery's depth of charge and discharge, as well as its charge and discharge power. These constraints are set as follows: ; ; In the formula: Let t be the total battery capacity during time period t. , The minimum and maximum energy storage capacity of the battery; , This represents the maximum charging and discharging power of the system's battery pack.

[0053] In practice, since the operating voltage of supercapacitors has a certain range, supercapacitor banks have maximum and minimum capacity constraints, as follows.

[0054] ; ; In the formula: , The minimum and maximum energy storage of a supercapacitor bank; for The capacity of the supercapacitor bank at any given time; The minimum power required for charging and discharging a supercapacitor; The maximum power for charging and discharging a supercapacitor.

[0055] Power balance constraints: ; Maximum power constraints for wind turbines and photovoltaic modules: ; ; Furthermore, the parameters (α, K) of the VMD algorithm for decomposing the hybrid energy storage output in the inner layer described in step S3 are optimized using a hybrid optimization algorithm DE_WOA that combines the differential evolution algorithm and the whale optimization algorithm. Its characteristics are described as follows: (1) Optimize the objective function (α, K): This embodiment uses the minimum sample entropy (SE) as the fitness function to optimize the VMD parameters. Sample entropy can effectively evaluate the disorder of each component IMF. For a time series with data of type K, SE is calculated as follows: ; Where K is the number of VMD decomposition layers (a parameter to be optimized). For the first One modal component, Indicates the first The sample entropy of an IMF.

[0056] (2) DE_WOA Algorithm: In the proposed two-layer optimization model, the outer layer focuses on capacity configuration of distributed power sources and hybrid energy storage systems, while the inner layer concentrates on optimizing variational mode decomposition (VMD) parameters. In actual computation, the outer layer optimization is typically set to 500 fitness evaluation calculations. If the inner layer VMD parameter optimization uses too many calculations, it will require hundreds of thousands of mode decomposition operations, significantly increasing the overall computation time. Therefore, obtaining the optimal VMD parameters (penalty factor α and mode number K) with fewer iterations and a smaller population size is crucial for improving the efficiency and accuracy of the system capacity configuration model. This embodiment proposes a differential evolution algorithm combined with a whale optimization algorithm to optimize VMD parameters, effectively solving this problem and improving the convergence speed and solution quality of the entire optimization model.

[0057] The DE_WOA algorithm used in this embodiment can be represented as follows: ,,in , , For population size, The maximum number of iterations, To solve for the variable dimension, the basic idea of ​​the DE algorithm is to first generate experimental individuals through mutation and crossover operators, and then generate a new generation of superior individuals from the parent individuals and experimental individuals through a greedy selection operator, eliminating inferior individuals. This allows the population to remember the optimal solution of each individual and share information within the population, thereby ensuring that the algorithm approaches the optimal solution. This embodiment uses a variant of DE combined with WOA. First, let's introduce DE.

[0058] (1) Mutation operator: The mutation operator is the core of DE, and its purpose is to provide each target individual with... Generate a mutation vector The mutation strategy is as follows: ; In the formula, Not equal to And all distinct integers; This is a scaling factor used to control the scaling degree of the difference vector.

[0059] ; (2) Crossover operator: Generate mutation vectors using mutation operators. Then, the mutation is used to target individuals. and mutation vector Crossover is performed to generate experimental individuals. .

[0060] ; In the formula, The crossover probability; A random number uniformly distributed between [0,1]. It is a random integer between [1, D], used to ensure that at least one dimension of the experimental individual comes from the mutation vector.

[0061] (3) Selection operator: Test individuals to be generated Finally, the target individual is compared using a "greedy" selection operator. and test individuals This ensures that individuals with better fitness enter the next generation of the population, thereby ensuring the consistency of the evolutionary process.

[0062] ; In the formula, This represents the objective function value.

[0063] The DE_WOA algorithm proposed in this embodiment is a hybrid algorithm at the intersection of the DE algorithm. It incorporates the random search, prey encirclement, and spiral update strategies of the WOA algorithm into the intersection part of the DE algorithm. The hybrid crossover operator is as follows: ; ; The WOA update parameters are as follows.

[0064] ; The pseudocode for the DE_WOA algorithm is shown below: Furthermore, the energy and frequency joint clustering method described in step S4 performs high- and low-frequency segmentation, and its characteristics are described as follows: The preset output of the HESS to compensate for the microgrid deficit is calculated using the following formula: ; Let the input be The set of IMF components after HOA-VMD decomposition is as follows: ; Each line Indicates the first Each component, total There are 1 component, each component has a length of 1. .

[0065] In Hess Energy Storage (HESS), the battery, as an energy storage device, primarily handles the low-frequency portion of the system and most of the energy in the HESS, while the supercapacitor, as a power storage device, primarily handles the high-frequency portion. Determining the allocation between high and low frequencies is crucial. If the allocation point is too low, the supercapacitor will bear a larger share of the energy, increasing system costs; conversely, if the allocation point f is too high, the battery's charging and discharging power will increase, also raising system costs. Furthermore, most literature relies on subjective judgment in allocation strategies, lacking adaptability. Since this embodiment simultaneously optimizes the system's wind and solar installed capacity, the output is uncertain during the iteration process; therefore, subjectively judging the frequency allocation point is unsuitable for this problem.

[0066] Therefore, a high- and low-frequency allocation method based on Hilbert and energy joint clustering is proposed. This method avoids the shortcomings of traditional subjective judgment. Furthermore, this method can simultaneously consider the rate of change and amplitude intensity of the signal, achieving a more reasonable high- and low-frequency mode division, avoiding the problem of insufficient supercapacitor load allocation, and providing a more reliable input basis for subsequent capacity optimization of energy storage systems.

[0067] For the IMF components After performing the Hilbert transform, an analytic signal is constructed: ; in This is the Hilbert transform.

[0068] Calculate the phase: ; Instantaneous frequency is calculated using the difference approximation: ; In actual calculations, negative frequencies are removed, and the average instantaneous frequency is taken as the representative frequency of the IMF. ; The energy (RMS) of each IMF component is: ; Each IMF is characterized as a joint feature: ; All IMF features constitute a data matrix: ; K-means clustering is used for high- and low-frequency classification (K=2): ; Let the high-frequency IMF index set be The low-frequency set is The preset signals for the battery and supercapacitor are obtained: ; In the formula, , , These are the preset outputs of the system's battery, supercapacitor, and hybrid energy storage, respectively, and res is the decomposed residual.

[0069] However, in actual operation, due to the limitations of HESS's own capacity, the energy storage output needs to be adjusted. The adjusted power is as follows: Battery charging and discharging power correction: ; Supercapacitor charging and discharging power correction: ; ; in, , , This represents the actual output of the modified battery, supercapacitor, and hybrid energy storage system. Figure 3A flowchart for high and low frequency power allocation.

[0070] Furthermore, the characteristics of the wind and solar curtailment rate, load shortage rate, and hybrid energy storage charging and discharging power output mentioned in step S5 are described as follows: The rated power of this invention is calculated as follows: based on the corrected actual power, the power configuration capacity of the HESS is obtained according to the corrected actual output power of the HESS: ; ; In the formula, , These are the rated power of the battery and the supercapacitor, respectively.

[0071] The calculation process for wind and solar curtailment rate and load power shortage rate is as follows: Figure 4 .

[0072] This invention uses a case study of annual wind speed, light intensity, and load in an industrial park in Xinjiang, with a time resolution of 15 minutes. The k-means algorithm is used to cluster the data for all four seasons throughout the year. The elbow method and silhouette coefficient method are employed to determine the number of clusters. Weighted summation is then performed on each typical day to obtain the final typical day data for each season. (See Error! Reference source not found.)

[0073] The distributed power generation parameters selected in this invention are shown in Table 1. Table 2 shows the typical daily values ​​of α and K for VMD obtained after optimization by the DE_WOA algorithm in the inner layer.

[0074] Table 1 Table 2 Figure 6 , Figure 7 , Figure 8 , Figure 9 Table 3 shows the VMD decomposition diagrams for spring, summer, autumn, and winter. Based on each component, the energy and average frequency of each IMF are calculated. Then, according to the energy-frequency joint clustering method mentioned in the text, the distribution data of high and low frequency IMFs are obtained, as shown in Table 3. The capacity configuration results for typical days in each season are shown in Table 4. Based on the Hilbert spectrum diagrams of each typical day, the energy-frequency joint clustering algorithm proposed in this embodiment can effectively allocate high and low frequency power. Figure 12 The spectrum diagrams of VMD decomposition on typical days in each season are analyzed. Based on the spectrum diagrams, the proposed energy plus frequency joint clustering effectively distinguishes between low and high frequencies. Low frequencies are responsible for most of the energy, while high frequencies are responsible for a small portion of the system's energy.

[0075] Table 3 Table 4 Figure 10 The figures show the output curves of each component within the system on typical days in spring, summer, autumn, and winter, as well as the energy waste and load interruption. The battery output curve is smooth, while the capacitor curve fluctuates more in spring, further illustrating the accuracy of the high-low frequency allocation proposed in this embodiment. Figure 11 The SOC curves of batteries and supercapacitors represent typical days in spring, summer, autumn, and winter. Batteries show a smooth and slow change in SOC, which is beneficial for extending battery life. Supercapacitors, on the other hand, exhibit a more rapid fluctuation in their SOC curves and primarily handle the high-frequency components of the system.

[0076] This embodiment uses the annual wind speed, solar irradiance, and load data of an industrial park in Xinjiang for case analysis, with a time resolution of 15 minutes. The information is summarized, and typical daily data for each season are extracted. Based on the provided parameters, a two-layer VMD capacity configuration optimization model for an integrated "wind-solar-hybrid energy storage" microgrid is established. The hybrid energy storage consists of lithium iron phosphate batteries and supercapacitors. The outer layer optimizes the capacity configuration of wind turbines, photovoltaic modules, and hybrid energy storage, while the inner layer decomposes the system deficit signal and allocates it to high and low frequencies. The CPLEX solver is used to solve the outer layer model and calculate relevant reliability indicators. The DE_WOA hybrid optimization algorithm, combining differential evolution and whale optimization algorithms, is used to optimize the VMD (variable mode decomposition) parameters (α, K) used for high and low frequency signal decomposition in the inner layer. Finally, under the premise of satisfying the corresponding constraints, the configuration is optimized for typical days in each of the four seasons. This invention addresses a nonlinear, discrete optimization problem—the capacity configuration optimization of an integrated microgrid combining hydropower, wind power, and hybrid energy storage—by employing the CPLEX solver to solve the outer layer model. The total capacity of wind turbines, photovoltaic units, lithium iron phosphate batteries, and supercapacitors are used as decision variables. The inner layer function utilizes a hybrid algorithm combining differential evolution and whale optimization to optimize the VMD algorithm parameters. Finally, with system power availability and related operational requirements as constraints, a mathematical model is established with the objective function of minimizing the typical daily lifecycle cost and daily operating cost, yielding the final microgrid capacity configuration result. This method exhibits high accuracy and stability, providing guidance for accelerating the rapid construction of microgrids in practical engineering. This invention comprehensively considers the annual wind and solar curtailment rate and system deficit rate, requiring only wind and solar data and load data for the microgrid construction environment, making it low-cost and easy to implement in practical applications.

[0077] In summary, the hybrid energy storage microgrid capacity configuration method proposed in this invention, based on two-layer VMD optimization decomposition, can allocate the hybrid energy storage component to high and low frequencies, balance the daily wind and solar curtailment rate and the daily deficit rate, thereby obtaining the optimal distribution terminal layout planning scheme. This method has certain guiding significance for accelerating the rapid construction of microgrids in practical engineering.

[0078] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for configuring the capacity of a hybrid energy storage microgrid based on two-layer optimization decomposition, characterized in that, Includes the following steps: Acquire the annual wind and solar resource data and load data of the target area, and extract several typical daily data based on the wind and solar resource data and load data; Based on the typical daily data, a two-layer capacity configuration optimization model is constructed with the goal of minimizing the overall cost of the microgrid. The outer layer of the two-layer capacity configuration optimization model is used to optimize the installed capacity of distributed power sources and hybrid energy storage, while the inner layer is used to decompose the system power deficit signal and perform high and low frequency allocation. A hybrid algorithm combining differential evolution and whale optimization is used to optimize and solve the variational mode decomposition algorithm used for signal decomposition in the inner layer. The optimized variational mode decomposition algorithm is used to decompose the system power deficit signal, and the energy and frequency joint clustering method is used to adaptively divide the multiple modal components obtained by decomposition into high-frequency components and low-frequency components, so as to allocate them to the supercapacitor and the battery respectively. Based on the power allocation scheme, the outer capacity configuration problem of the capacity configuration two-layer optimization model is solved by an optimization solver, and the optimal capacity configuration results of each distributed power source and hybrid energy storage that meet the preset operating constraints are output.

2. The method according to claim 1, characterized in that, Acquire annual wind and solar resource data and load data for the target area, and extract several typical daily data based on the wind and solar resource data and load data, specifically including: Based on the time-series data of wind speed, light intensity and load acquired throughout the year, the data are classified by season; For each season's data, a clustering algorithm is used to determine the number of typical days for that season. The multiple candidate day data obtained from the clustering are then weighted and summed to finally obtain a single typical day data that represents the characteristics of the corresponding season.

3. The method according to claim 1, characterized in that, A two-level capacity configuration optimization model is constructed with the goal of minimizing the overall cost of microgrids. The objective function of the two-level capacity configuration optimization model is the typical daily comprehensive cost, which includes the annual system investment and installation cost, annual operation and maintenance cost, wind and solar curtailment penalty cost, load interruption compensation cost, equipment replacement cost, and additional energy curtailment penalty cost. Furthermore, the constraints that the capacity configuration dual-layer optimization model must satisfy include: the upper limit constraints on the number or capacity of wind and solar power generation units and energy storage units, the state of charge and charging and discharging power constraints of batteries and supercapacitors, system power balance constraints, and the maximum output constraints of wind and solar power generation units.

4. The method according to claim 1, characterized in that, A hybrid algorithm combining differential evolution and whale optimization is used to optimize the variational mode decomposition algorithm used for signal decomposition in the inner layer. Specifically, this includes: Based on the preset optimization objective, the parameter population of the hybrid algorithm is initialized, where each individual represents a set of candidate penalty factors and modality decomposition layers; Based on the current parameter population, new candidate parameter combinations are generated using the mutation and crossover strategies of the hybrid algorithm. Based on the combination of candidate parameters, variational mode decomposition is performed on the system power deficit signal, and the sum of sample entropy of each mode component after decomposition is calculated. Based on the sum of the entropies of each sample obtained by calculation, the parameter population is updated by selecting a strategy to retain individuals with better fitness. Based on the preset iteration termination condition, the above process of generating candidate parameters, calculating sample entropy and updating the population is repeated to finally obtain the optimal penalty factor and mode decomposition layer number that minimizes the sum of sample entropy.

5. The method according to claim 4, characterized in that, The preset optimization objective is to minimize the sum of sample entropies; specifically, calculating the sum of sample entropies involves: calculating the sample entropy of each modal component based on the time series obtained from variational mode decomposition, and summing the sample entropy values ​​of all components.

6. The method according to claim 4, characterized in that, The method of generating new candidate parameter combinations using the mutation and crossover strategy of the hybrid algorithm specifically includes: Based on the current population, the first mutation vector is generated using the mutation operator of the differential evolution algorithm; Based on the best individual in the current population and the prey encirclement and spiral update strategy of the whale optimization algorithm, a second mutation vector is generated. Based on the preset crossover probability, the first mutation vector and the second mutation vector are crossed with the target individual to generate experimental individuals as new candidate parameter combinations.

7. The method according to claim 1, characterized in that, A joint energy and frequency clustering method is used to adaptively divide the multiple modal components obtained from the decomposition into high-frequency components and low-frequency components, specifically including: Calculate the average instantaneous frequency and energy characteristics of each modal component to form a joint feature vector for each component; Based on the joint feature vector of all modal components, a clustering algorithm is used to divide all modal components into two clusters; One cluster is identified as a high-frequency component and assigned to the supercapacitor; the other cluster is identified as a low-frequency component and assigned to the battery. After the component allocation is completed, the preset output signal is further modified according to the rated capacity and charging / discharging power limits of the battery and supercapacitor to obtain the actual executable charging / discharging power command.

8. The method according to claim 1, characterized in that, Based on the power allocation scheme, an optimization solver is used to solve the outer capacity configuration problem of the two-layer capacity configuration optimization model, outputting the optimal capacity configuration results for each distributed power source and hybrid energy storage that satisfy preset operating constraints. Specifically, this includes: Based on the power allocation scheme, the capacity configuration bi-level optimization model is solved using an optimization solver. The total capacity of wind turbines, photovoltaic modules, batteries, and supercapacitors are used as decision variables. Under the condition of satisfying all preset constraints, the capacity configuration scheme that minimizes the comprehensive cost on a typical day is obtained. Based on the capacity configuration scheme and the power allocation scheme obtained from the solution, the wind curtailment rate and the load shortage rate of the system under this configuration are calculated. The output of the capacity configuration scheme, the corresponding hybrid energy storage power configuration, and the wind curtailment rate and load power shortage rate are used as the optimal capacity configuration result.

9. An electronic device comprising a memory, a processor, and a computing program stored in the memory and executable on the processor, characterized in that, When the processor executes the computing program, it implements the method of any one of claims 1-8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-8.

Images